Inspiration
Imagine walking onto a construction site.
Hundreds of workers move simultaneously. Cranes swing overhead, forklifts haul materials across the yard, concrete pours on one floor while electrical work continues on another. Every worker is making dozens of decisions a minute — where to walk, what tool to grab, what task comes next, whether an area is safe, whether a critical checklist item was actually completed.
And yet the site itself remembers none of it.
When a worker installs a beam, only that worker knows it happened. When materials get moved, everyone else wastes time hunting for them. When a supervisor finishes an inspection, the rest of the crew may still not know the area is clear. When an experienced worker spots a mistake, that knowledge often stays in their head instead of reaching whoever needs it next.
Today's AI can detect helmets, identify machinery, flag hazards — but it still treats every camera frame as an isolated moment. It sees, but it doesn't understand. It can't answer questions like:
- What task is this worker actually performing?
- Did they skip a critical workflow step?
- Who nearby can help them right now?
- Has this hazard already been reported?
- Is this area actually ready for the next phase?
We believed construction sites deserved more than surveillance — they deserved collective intelligence.
That became Kairos.
Kairos combines smart glasses, a custom ESP32-S3 haptic wristband, ultra-wideband (UWB) positioning, edge AI, and a semantic digital twin into one shared understanding of the site. UWB localization means Kairos doesn't just know what happened — it knows where. Every worker, tool, material, and hazard becomes part of a spatial memory that keeps evolving, turning isolated observations into real context.
The wristband turns AI from something workers have to look at into something they can feel. Directional haptic feedback steers them away from hazards, nudges them about missed steps, gives them a one-touch SOS, and works even where machinery noise would drown out any voice prompt.
Most importantly, Kairos treats workers as a connected team, not isolated individuals. If someone hesitates mid-task, the system can identify a nearby coworker with the right experience and connect them — a kind of human swarm intelligence where knowledge moves across the site instead of staying trapped in one person's head.
By continuously observing, understanding, and remembering what happens on site, Kairos builds a living memory that supports workers, supervisors, and project managers alike — anticipating problems and guiding people in real time, instead of just documenting the aftermath.
We didn't want to build another AI that watches construction sites. We wanted to build one that understands them.
What it does
Kairos is an AI-native cognitive copilot that helps workers in real time while giving managers a live, ground-truth view of everything happening on site.
Workers wear smart glasses and a haptic wristband through their shift. As they move around, Kairos continuously observes their surroundings, understands what task they're performing, tracks workflow progress, and flags hazards or missed procedures — proactively, not after the fact. It can nudge a worker who skipped a step, warn them as they approach a danger zone, guide them around it with directional haptics, or connect them with a nearby coworker who has the experience to help.
Every meaningful observation feeds a living digital twin of the site — a continuously updated record of workers, materials, equipment, completed tasks, inspections, and hazards, with zero manual reporting.
Managers see all of this through a live 3D dashboard built in Three.js. Instead of scrubbing through camera feeds, they get an interactive model of the site showing:
- Live worker locations
- Current task progress
- Workflow completion status
- Hazard zones and active alerts
- Equipment and material locations
- Real-time project progress
- A historical event timeline
Workers can also just talk to Kairos:
- "What's my next task?"
- "Can I start work in this zone?"
- "Where are the remaining steel rods?"
- "Who nearby can help me?"
By pairing real-time guidance for workers with a living digital twin for managers, Kairos turns construction sites from places that just record what happened into places that understand the work, retain knowledge, and catch mistakes before they happen.
Built With
- computer-vision
- edge-computing
- esp32
- fastapi
- langgraph
- machine-learning
- mediapipe
- mqtt
- opencv
- python
- pytorch
- react
- tensorrt
- three.js
- uwb
- websockets
- yolo
Log in or sign up for Devpost to join the conversation.